Arxiv:1804.09082V3 [Astro-Ph.SR] 31 May 2019 Upr O H Oeaceinmcaimo Lntformation
Total Page:16
File Type:pdf, Size:1020Kb
Draft version June 4, 2019 Typeset using LATEX manuscript style in AASTeX61 RETIRED A STARS REVISITED: AN UPDATED GIANT PLANET OCCURRENCE RATE AS A FUNCTION OF STELLAR METALLICITY AND MASS Luan Ghezzi,1 Benjamin T. Montet,2, ∗ and John Asher Johnson3 1Observat´orio Nacional, Rua General Jos´eCristino, 77, 20921-400, S˜ao Crist´ov˜ao, Rio de Janeiro, RJ, Brazil; [email protected] 2Department of Astronomy and Astrophysics, University of Chicago, 5640 S. Ellis Ave, Chicago, IL 60637, USA 3Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 USA ABSTRACT Exoplanet surveys of evolved stars have provided increasing evidence that the formation of giant planets depends not only on stellar metallicity ([Fe/H]), but also the mass (M⋆). However, measuring accurate masses for subgiants and giants is far more challenging than it is for their main-sequence counterparts, which has led to recent concerns regarding the veracity of the correlation between stellar mass and planet occurrence. In order to address these concerns we use HIRES spectra to perform a spectroscopic analysis on an sample of 245 subgiants and derive new atmospheric and physical parameters. We also calculate the space velocities of this sample in a homogeneous manner for the first time. When reddening corrections are considered in the calculations of stellar masses and a -0.12 M⊙ offset is applied to the results, the masses of the subgiants are consistent with their space velocity distributions, contrary to claims in the literature. Similarly, our measurement of their arXiv:1804.09082v3 [astro-ph.SR] 31 May 2019 rotational velocities provide additional confirmation that the masses of subgiants with M⋆ ≥ 1.6 M⊙ (the “Retired A Stars”) have not been overestimated in previous analyses. Using these new results for our sample of evolved stars, together with an updated sample of FGKM dwarfs, we confirm that giant planet occurrence increases with both stellar mass and metallicity up to 2.0 M⊙. We show that the probability of formation of a giant planet is approximately a one-to-one function of the [Fe/H] total amount of metals in the protoplanetary disk M⋆ 10 . This correlation provides additional support for the core accretion mechanism of planet formation. 2 Ghezzi, Montet & Johnson Keywords: stars: fundamental parameters — stars: atmospheres — stars: evolution — planets: formation 1. INTRODUCTION The ever increasing sample of exoplanets has facilitated many robust studies of how planet formation and evolution are affected by the physical properties of their host stars. One such result is the correlation between stellar metallicity and the occurrence rate of giant planets around FGK dwarf and subgiant stars (e.g., Gonzalez 1997; Fischer & Valenti 2005; Ghezzi et al. 2010a; Wang & Fischer 2015). Similar analyses for the stellar masses, however, are much more scarce and controversial. The main limitation appears on the massive end (&1.2 M⊙) of the planet search samples, since the detection or confirmation of exoplanets around main-sequence A and early F stars using radial velocities (RVs) is more difficult due to their higher effective temperatures and larger rotational velocities. In order to overcome this issue, radial velocity surveys have focused on intermediate-mass stars. As stars evolve off of the main sequence towards the red giant branch (RGB), their effective temperatures decrease and their rotation velocities slow down, resulting in spectra from which precise Doppler shifts can be measured. To date, more than 100 exoplanets have been detected around more than 1000 evolved stars (Jofr´eet al. 2015; Niedzielski et al. 2015). In one of these efforts, Johnson et al. (2010a) analyzed a sample of 246 subgiants from the SPOCS IV catalog, as well as 948 FGKM dwarfs and subgiants from the Keck M Dwarf Survey (Butler et al. 2006) and the original SPOCS catalog (Valenti & Fischer 2005). Using the complete sample of 1194 stars, with masses ranging from 0.2 M⊙to 2.0 M⊙, Johnson et al. (2010a) determined that the occurrence rate of giant planets increases approximately linearly with stellar mass, from ∼3% for M dwarfs, to ∼8% for FGK dwarfs, ∗ Sagan Fellow Retired A Stars Revisited 3 to ∼14% for stars more massive than 1.5 M⊙, the so-called ”Retired A stars.” This correlation appears to be supported, at least for the FGKM dwarfs, by results from transit surveys (Fressin et al. 2013; Gaidos et al. 2013). Johnson et al. (2010a) also confirmed that the planet-metallicity correlation holds for both M dwarfs and the Retired A Stars. This suggests that planet formation is significantly affected by the stellar mass and chemical composition—two stellar properties that provide crucial links to the physical properties of stars’ protoplanetary disks. These findings were questioned by Lloyd (2011), who pointed out that some discrepancies between the rotational velocities distributions of evolved planet hosts and field stars could be traced back to erroneous mass determinations. Moreover, Lloyd found that the numbers of massive subgiants (M⋆& 1.5 M⊙) observed by Johnson et al. (2010a) seems inconsistent with the predictions from Galactic stellar population models. More massive stars are formed less frequently according to the initial mass function (IMF) and also have a relatively rapid evolution across the Hertzprung Gap. Based on these arguments, Lloyd concludes that many of the Retired A Stars should in fact lie in the mass range 1.0–1.2 M⊙. The errors on the original masses could stem from systematic uncertainties on spectroscopically determined atmospheric parameters or ambiguities in the predictions from stellar evolution models. Johnson et al. (2013) addressed the second concern by generating Galactic stellar population models with TRILEGAL (Girardi et al. 2005). Using a magnitude-limited sample (instead of the volume- limited sample used by Lloyd 2011), they produced a simulated distribution that allows for a larger number of massive stars than Lloyd found—an effect similar to the Malmquist bias—and is thus consistent with the masses determined by Johnson et al. (2010a). Lloyd (2013), on the other hand, argues that the mass distribution is not sensitive to the Malmquist bias, but instead to the usage of different Galactic models or input parameters within a given model. Using the Besan¸con model (Robin et al. 2003), Lloyd (2013) obtains a mass distribution with less massive subgiants that can not be reconciled with the observed data from Johnson et al. (2010a). The fraction of massive subgiants within the Retired A Star sample was also questioned by Schlaufman & Winn (2013). Based on an analysis that is independent of stellar evolution mod- 4 Ghezzi, Montet & Johnson els, they showed that the space velocity dispersion of the subgiants with planets were larger than those of a sample of main-sequence A5–F0 stars, but consistent with the that obtained for F5–G5 dwarfs. They thus conclude that the subgiants with planets are most likely the evolved counterparts of less massive main-sequence solar-type stars, instead of “Retired A Stars”. Their analysis also revealed that no more than 40% of the planet-hosting subgiants could have been A5–F0 stars more massive than ∼1.3 M⊙ while on the main sequence. In another study, Sousa et al. (2015b) pointed out that, while the masses they determined for planet-host stars were in general good agreement with values from the literature, there were some notable exceptions. In particular, differences of up to 100% were found for a few evolved stars. Note, however, that some of these inconsistencies are explained by erroneous mass determinations in the original papers, as explained by (Takeda & Tajitsu 2015, see Section 4.2). More recent studies of the stellar mass dependence of planet occurrence have been performed by Reffert et al. (2015) and Jones et al. (2016). In the former study, the analysis of 373 G and K giants reveals that giant planet occurrence increases with stellar mass in the range 1.0 – 1.9 M⊙, consistent with the findings of Johnson et al. (2010a). Moreover, there seems to be a maximum for +0.1 the occurrence rate as a function of stellar mass at 1.9−0.5 M⊙ and a rapid decrease for more massive stars. The latter study uses a sample of 166 giant stars and finds a consistent result: occurrence rate increases with stellar mass, with the maximum at 2.1 M⊙. In the previous works mentioned, masses for the evolved stars were estimated by comparing ob- served properties (effective temperature, metallicity, and luminosity or surface gravity) with grids of stellar evolution models (e.g., Bressan et al. 2012). Although the accuracy of this method has been thoroughly tested for main-sequence stars (e.g., Torres et al. 2010), results for sugbiants and giants are still debated, as is clear from the above discussion. In an additional effort to test if stellar evolution models are also reliable for evolved stars, Ghezzi & Johnson (2015) showed that PARSEC evolutionary tracks (Bressan et al. 2012) coupled with the code PARAM (da Silva et al. 2006) can recover model-independent masses from eclipsing binaries and asteroseismology within ∼4% for the interval ∼0.7 – 4.5 M⊙. Therefore, the method itself does not present any issues that would signif- Retired A Stars Revisited 5 icantly overestimate the masses of subgiants and giants. Nevertheless, incorrect input parameters could still lead to erroneous mass determinations. To avoid model-dependent issues, most recent efforts to solve the Retired A Stars controversy focused on the determination of empirical masses. Johnson et al. (2014) performed multiple mass measurements for the bright giant HD 185351 using spectroscopic, interferometric and asteroseismic data. Although discrepancies on the 2.6σ level were observed, all determined values confirm that HD 185351 is more massive than 1.5 M⊙.